Location: Location not imported yet.Title: Land Cover Characterization for Hydrological Modeling Using Thermal Infrared Emissivities) Author
Submitted to: International Journal of Remote Sensing
Publication Type: Peer reviewed journal
Publication Acceptance Date: 11/8/2009
Publication Date: 8/4/2010
Citation: French, Andrew N. and Inamdar, Anand(2010) Land cover characterization for hydrological modelling using thermal infrared emissivities, International Journal of Remote Sensing, 31: 14, 3867 — 3883 Interpretive Summary: Water consumed by crops changes by large amounts throughout a growing season, and measuring these amounts is vital for water conservation and sustainable agriculture. One way to measure water accurately is to develop satellite-based remote sensing tools, which can give overviews of crops. However, these views often mistake crop as soil, which means that water use estimates at these times can be inaccurate. The difficulty occurs because remote sensing vegetation indices cannot easily distinguish between brown vegetation and bare soil surfaces. A method has been developed that will reduce the problem by observing the land in multiple thermal infrared wavelengths and producing maps of thermal efficiency, otherwise known as emissivity. These maps can distinguish soil from vegetation, whether green or not, and can be used with other remote sensing data sets to improve water consumption estimates. The approach was tested for observations over winter wheat fields grown in the Southern Great Plains in 2007 and accurately detected plant senescence and harvest times. Using emissivities suggests that water consumed during these times is about 1.5 mm/day less than would otherwise be predicted. This result will be useful for agricultural and hydrological scientists investigating better ways to monitor the water cycle.
Technical Abstract: Remote sensing with multispectral thermal infrared observations has the potential to improve regional scale estimation of evapotranspiration (ET) by constraining the land surface energy balance in a way that is not possible using more conventional remote sensing techniques. Current models use data from visible and near infrared bands to obtain vegetative cover estimates, and sometimes also use data from thermal infrared bands to obtain land surface temperature estimates. Together these data sets have the potential to yield good regional estimates of ET. However it may be possible to retrieve even better ET estimates using remotely sensed thermal infrared emissivity, which is a surface property that provides information about fractional vegetative cover and is independent of plant greenness. Thus thermal emissivity data, when used with a vegetation index, can improve discrimination between senescent vegetation, green vegetation, and bare soil surfaces. To demonstrate this potential, emissivities derived from clear-sky MODIS observations obtained in 2007 over the Southern Great Plains study areas (Oklahoma and Kansas, USA) were compared with changes in NDVI for two land cover types, winter wheat and grazingland. For all regions emissivity changes were independent of NDVI, indicating its sensitivity to standing canopies, regardless of growth stage or senescence. Thus emissivity data were shown to be seasonally dynamic, able to detect wheat harvest timing, and helpful for modeling ET. The effect of incorporating emissivities into a surface energy balance model could be significant because the approach reduces net radiation at the soil surface compared with more conventional vegetation index methods.